Methods, systems, and apparatus, including computer programs encoded on computer storage media, for scoring candidate routes. One of the methods includes obtaining a predictive model trained on training examples from trip log data, wherein each training example has feature values from a particular trip and a value of a dependent variable that represents an outcome of a portion of the particular trip, wherein the features of each particular trip include values obtained from one or more external data feed sources that specify a value of a sensor measurement at a particular point in time during the trip. Sensor values from one or more external data feed sources of a sensor network are received. Feature values are generated using the sensor values received from the one or more external data feed sources. A predicted score is computed for each route using the feature values for the candidate route.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
3. The computer-implemented method of claim 2, wherein the corresponding waypoint-specific score represents a respective benefit value of arriving at the respective waypoint during a predicted time window indicated by the respective candidate route.
This invention relates to route optimization for autonomous or semi-autonomous vehicles, addressing the challenge of selecting the most efficient route based on dynamic conditions and waypoint-specific objectives. The method evaluates multiple candidate routes by calculating waypoint-specific scores that quantify the benefit of arriving at each waypoint within a predicted time window. These scores are derived from factors such as traffic conditions, time-sensitive tasks, or resource availability at each waypoint. The method then selects the optimal route by aggregating these scores, ensuring the chosen path maximizes overall efficiency or utility. The invention also includes generating candidate routes from a starting location to a destination, where each route includes multiple waypoints. The waypoint-specific scores are computed based on real-time or predicted data, allowing the system to adapt to changing conditions. This approach improves decision-making for autonomous navigation, logistics, or delivery systems by dynamically prioritizing routes that align with time-sensitive objectives or resource constraints. The method may also incorporate historical data or machine learning models to refine score predictions over time.
4. The computer-implemented method of claim 2, wherein the plurality of training samples is based on trip log data for previously-completed trips, and wherein the corresponding dependent score comprises, for each respective training waypoint of the particular trip, a waypoint-specific training score that indicates an extent to which a task was successfully performed at the respective training waypoint during a corresponding time window.
This invention relates to a computer-implemented method for training a machine learning model to evaluate task performance during trips, such as delivery or service routes. The method addresses the challenge of assessing whether tasks (e.g., deliveries, inspections) were completed correctly at specific waypoints along a route, using historical trip data to improve accuracy. The method processes trip log data from previously completed trips, where each trip includes multiple waypoints (locations where tasks are performed). For each waypoint in a training sample, a waypoint-specific score is generated, indicating how successfully a task was performed during a predefined time window. These scores are used to train a model that can later evaluate new trips by predicting task success at waypoints based on learned patterns from historical data. The training samples are derived from past trips, ensuring the model learns from real-world scenarios. The waypoint-specific scores provide granular feedback, allowing the model to distinguish between successful and unsuccessful task completions at different locations. This approach improves the reliability of task performance assessments in logistics, field service, or other route-based operations. The method enhances automation by reducing reliance on manual checks and improving decision-making for route optimization and task verification.
5. The computer-implemented method of claim 1, wherein each respective waypoint of the same predetermined plurality of waypoints is associated with a corresponding task to be completed at the respective waypoint, and wherein the corresponding score for the respective candidate route is based on a likelihood of successfully completing the corresponding task of at least one waypoint of the same predetermined plurality of waypoints.
This invention relates to route optimization for tasks, particularly in scenarios where multiple waypoints are involved, each associated with a specific task. The problem addressed is efficiently determining the best route through a set of waypoints while accounting for the likelihood of successfully completing tasks at each location. The method evaluates candidate routes by assessing the probability of task completion at one or more waypoints along the route. Higher-scoring routes are those where the tasks are more likely to be successfully completed, ensuring optimal task fulfillment. The system considers the inherent variability in task success rates at different waypoints, allowing for dynamic route adjustments based on real-time or predicted conditions. This approach is useful in logistics, delivery services, field operations, or any application where task completion reliability is critical to route selection. By incorporating task success likelihood into route scoring, the method improves efficiency and reliability compared to traditional distance- or time-based routing algorithms. The invention ensures that routes are not only optimal in terms of distance or time but also maximize the probability of successfully completing all assigned tasks.
6. The computer-implemented method of claim 1, wherein each respective candidate route of the plurality of candidate routes represents a proposed rerouting for subset of a trip of the vehicle.
This invention relates to dynamic route optimization for vehicle navigation systems, addressing the challenge of efficiently rerouting vehicles in real-time to avoid traffic congestion, road closures, or other disruptions. The method involves generating multiple candidate routes for a vehicle's trip, where each candidate route represents a proposed rerouting for a specific segment or subset of the trip. The system evaluates these candidate routes based on factors such as travel time, fuel efficiency, or other optimization criteria to select the most efficient path. The method may also incorporate real-time data, such as traffic conditions or weather updates, to dynamically adjust the candidate routes. Additionally, the system may consider constraints like vehicle type, cargo requirements, or driver preferences when generating and evaluating the routes. The goal is to improve overall trip efficiency, reduce delays, and enhance fuel economy by dynamically optimizing the vehicle's path throughout the journey. The invention is particularly useful for fleet management, logistics, and autonomous vehicle navigation systems.
9. The computer-implemented method of claim 1, wherein the independent feature values for the respective candidate route are based on sensor data received from one or more remote devices configured to generate one or more of vehicle data, passageway data, weather data, geopolitical data, facility data, or agent data.
This invention relates to a computer-implemented method for optimizing route selection using sensor data from remote devices. The method addresses the challenge of determining the most efficient route for a vehicle or agent by incorporating real-time data to assess route conditions dynamically. The method involves analyzing independent feature values for candidate routes, where these values are derived from sensor data collected from remote devices. These devices generate various types of data, including vehicle data (e.g., speed, position, or status of nearby vehicles), passageway data (e.g., road conditions, traffic congestion, or lane availability), weather data (e.g., precipitation, visibility, or temperature), geopolitical data (e.g., border crossings, political unrest, or regulatory changes), facility data (e.g., fuel stations, rest areas, or maintenance facilities), and agent data (e.g., status or location of other agents or entities involved in the route). By integrating this diverse sensor data, the method evaluates the suitability of each candidate route based on real-world conditions, enabling more accurate and adaptive route optimization. This approach improves decision-making for navigation systems, logistics planning, and autonomous vehicle operations by accounting for dynamic environmental and operational factors.
10. The computer-implemented method of claim 1, wherein the independent feature values for the respective candidate route represent one or more of agent-specific information, payload-specific information, vehicle-specific information, passageway-specific information, path-specific information, environment-specific information, facility-specific information, or customer-specific information.
This invention relates to a computer-implemented method for optimizing route selection in autonomous or semi-autonomous systems, particularly for agents such as drones, robots, or autonomous vehicles. The method addresses the challenge of efficiently determining optimal routes by incorporating diverse, independent feature values that influence route selection. These features include agent-specific information (e.g., capabilities, constraints, or status of the agent), payload-specific information (e.g., weight, size, or fragility of the payload), vehicle-specific information (e.g., power, speed, or sensor limitations), passageway-specific information (e.g., width, height, or obstacles in the path), path-specific information (e.g., distance, terrain, or traffic conditions), environment-specific information (e.g., weather, lighting, or hazards), facility-specific information (e.g., entry/exit points, loading zones, or restricted areas), and customer-specific information (e.g., delivery preferences, time constraints, or priority levels). By analyzing these features, the method dynamically evaluates candidate routes to select the most efficient or suitable path based on real-time or pre-existing data, improving decision-making in navigation and logistics applications. The approach enhances adaptability in dynamic environments, ensuring safer, faster, or more cost-effective routing.
11. The computer-implemented method of claim 1, wherein the machine learning model is configured to generate, for each respective path of one or more paths between waypoints in the corresponding ordering of the same predetermined plurality of waypoints specified by the candidate route, a predicted duration for the vehicle to travel the respective path, and wherein the corresponding score is based on the predicted duration for each respective path of the one or more paths.
This invention relates to route optimization for vehicles using machine learning. The problem addressed is efficiently determining optimal routes for vehicles by predicting travel durations along different paths between waypoints. The system uses a machine learning model to analyze multiple paths between waypoints in a predefined sequence, generating predicted travel durations for each path. These predictions are then used to calculate a score for the route, which reflects the efficiency or feasibility of the route based on the predicted durations. The model considers various factors that may affect travel time, such as traffic conditions, road types, or vehicle characteristics, to provide accurate duration estimates. By evaluating multiple paths between waypoints, the system can identify the most time-efficient or otherwise optimal route for the vehicle. This approach improves upon traditional routing methods by leveraging machine learning to dynamically assess travel conditions and optimize route selection. The invention is particularly useful in logistics, autonomous vehicle navigation, and fleet management, where precise travel time predictions are critical for efficiency and cost savings.
12. The computer-implemented method of claim 1, wherein the machine learning model is configured to generate, for each respective path of one or more paths between waypoints in the corresponding ordering of the same predetermined plurality of waypoints specified by the candidate route, a predicted distance to be traveled by the vehicle on the respective path, and wherein the corresponding score is based on the predicted distance for each respective path of the one or more paths.
This invention relates to route optimization for vehicles using machine learning. The problem addressed is efficiently determining optimal routes by predicting travel distances for different paths between waypoints, improving navigation systems and logistics planning. The method involves a machine learning model that processes a candidate route defined by a predetermined sequence of waypoints. For each possible path between consecutive waypoints in the route, the model predicts the distance the vehicle will travel. The system then calculates a score for the candidate route based on these predicted distances, allowing comparison of different route options. This approach accounts for variations in path distances due to factors like traffic, road conditions, or vehicle dynamics, enabling more accurate route selection. The machine learning model is trained to consider multiple paths between waypoints, not just direct routes, to assess all feasible travel options. By integrating predicted distances into route scoring, the system improves upon traditional methods that rely solely on static map data or simple heuristics. This enhances efficiency in applications such as autonomous vehicle navigation, delivery route planning, and fleet management. The invention provides a data-driven approach to route optimization, reducing travel time and fuel consumption by selecting paths with the most accurate predicted distances.
15. The system of claim 14, wherein the corresponding waypoint-specific score represents a respective benefit value of arriving at the respective waypoint during a predicted time window indicated by the respective candidate route.
This invention relates to route optimization systems for autonomous vehicles or navigation systems. The problem addressed is efficiently determining optimal routes by evaluating waypoints along candidate routes to maximize benefits, such as time savings, fuel efficiency, or other performance metrics. The system generates candidate routes between a starting location and a destination, each route comprising multiple waypoints. For each waypoint, the system calculates a waypoint-specific score representing the benefit of arriving at that waypoint within a predicted time window. The time window is derived from the candidate route's estimated arrival time at the waypoint. The system then selects the optimal route based on these scores, ensuring that the chosen path maximizes the cumulative benefit across all waypoints. The system may also account for dynamic factors such as traffic conditions, road closures, or vehicle constraints to refine the predicted time windows and scores. By dynamically adjusting the evaluation of waypoints, the system improves route selection accuracy and adaptability in real-world scenarios. The invention enhances navigation efficiency by prioritizing routes that offer the highest cumulative benefits, such as reduced travel time or improved energy consumption.
16. The system of claim 14, wherein the plurality of training samples is based on trip log data for previously-completed trips, and wherein the corresponding dependent score comprises, for each respective training waypoint of the particular trip, a waypoint-specific training score that indicates an extent to which a task was successfully performed at the respective training waypoint during a corresponding time window.
This invention relates to a system for evaluating task performance during trips, particularly in logistics or delivery operations. The system addresses the challenge of assessing how well tasks are completed at specific waypoints along a trip route, such as package deliveries, inspections, or service calls. The system uses historical trip log data from previously completed trips to generate training samples, each representing a waypoint along a route. For each waypoint in a training sample, a waypoint-specific training score is calculated, reflecting the success of task completion within a defined time window. This score quantifies factors like timeliness, accuracy, or completeness of the task. The system leverages these scores to train models or algorithms that can predict or evaluate task performance in future trips. By analyzing historical data, the system helps identify patterns, optimize routes, or improve operational efficiency. The invention focuses on granular, location-specific performance metrics to enhance decision-making in trip-based workflows.
17. The system of claim 13, wherein each respective waypoint of the same predetermined plurality of waypoints is associated with a corresponding task to be completed at the respective waypoint, and wherein the corresponding score for the respective candidate route is based on a likelihood of successfully completing the corresponding task of at least one waypoint of the same predetermined plurality of waypoints.
This invention relates to route optimization systems for autonomous or semi-autonomous vehicles, particularly those designed to navigate through a series of predefined waypoints while performing tasks at each location. The problem addressed is ensuring efficient and reliable task completion along a route, where the success of each task depends on factors like environmental conditions, vehicle capabilities, or other dynamic variables. The system evaluates multiple candidate routes between a starting point and a destination, each consisting of the same set of waypoints. At each waypoint, a specific task must be completed, such as picking up an object, scanning an area, or interacting with another system. The system calculates a score for each candidate route based on the likelihood of successfully completing the tasks at one or more waypoints along that route. This likelihood assessment considers factors like the vehicle's ability to reach the waypoint, the feasibility of performing the task upon arrival, and external conditions that may affect task success. By prioritizing routes with higher success probabilities, the system optimizes for both efficiency and reliability, ensuring that critical tasks are completed even in uncertain or dynamic environments. The approach is particularly useful in logistics, surveillance, or inspection applications where task completion is as important as reaching the destination.
18. The system of claim 13, wherein the independent feature values for the respective candidate route are based on sensor data received from one or more remote devices configured to generate one or more of vehicle data, passageway data, weather data, geopolitical data, facility data, or agent data.
This invention relates to a system for evaluating and selecting candidate routes based on sensor data from remote devices. The system addresses the challenge of optimizing route selection by incorporating real-time, multi-source data to improve decision-making for navigation or logistics applications. The system processes independent feature values for each candidate route, which are derived from sensor data collected by remote devices. These devices generate various types of data, including vehicle data (e.g., speed, location, or status), passageway data (e.g., road conditions or traffic flow), weather data (e.g., precipitation or temperature), geopolitical data (e.g., border restrictions or political events), facility data (e.g., fuel availability or maintenance schedules), and agent data (e.g., user preferences or historical behavior). By analyzing this diverse dataset, the system assesses route feasibility, efficiency, and safety, enabling dynamic adjustments to routing decisions. The system integrates these feature values to generate a comprehensive evaluation of each candidate route, allowing for optimized path selection based on real-world conditions. This approach enhances traditional routing algorithms by incorporating contextual and situational awareness, improving reliability and adaptability in dynamic environments. The use of remote sensor data ensures that the system remains responsive to changing conditions, supporting applications in autonomous vehicles, fleet management, or emergency response systems.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
June 28, 2023
June 4, 2024
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.